Intelligent Engineering Systems through Artificial Neural Networks Volume 18
21 Neural Network and Genetic Programming in Pressure Loss Estimation in Eccentric Pipe Flow
Download citation file:
- Ris (Zotero)
- Reference Manager
Studies of fluid flow in annular pipes have been popular in the petroleum engineering research. Most of the work has concentrated on CFD (Computational Fluid Dynamics) simulations, analytical and empirical models. In this study a neural network and evolutionary programming approach is developed to model the behavior of fluid flow in eccentric pipes. The model uses the fluid rheological parameters, density, mass flow rate, eccentricity, inner and outer pipe diameters, and predicts the pressure drop (ΔP) in the pipe in the flow direction. The evolutionary programming model uses basic mathematical operators, logarithm and sine functions. The results are compared with some experimental data obtained in literature and some Matlab CFD simulations. Preliminary studies indicate the neural network model performed better than the other models, evolutionary programming model can predict comparable pressure drop results, but not as effectively as the other models.